Artificial neural networks: Review

Yazici A. C. , Oegues E., Ankarali S., Canan S., Ankarali H., Akkus Z.

TURKIYE KLINIKLERI TIP BILIMLERI DERGISI, vol.27, no.1, pp.65-71, 2007 (Peer-Reviewed Journal) identifier identifier

  • Publication Type: Article / Article
  • Volume: 27 Issue: 1
  • Publication Date: 2007
  • Journal Indexes: Science Citation Index Expanded, Scopus
  • Page Numbers: pp.65-71
  • Keywords: neural networks, nonlinear dynamics, neurons, brain, classification, LINEAR-REGRESSION, TIME-SERIES, SOM


Artificial neural networks (ANNs) are computer softwares that were developed by simulating the working mechanism of human brain to accomplish the basic functions of the brain. ANNs have capability to learn, remember and then generalize the data to produce new information, and to detect the relationships between variables. There are considerable relations between the statistical methods and the neural networks. In the present study, biological neural network and neurons of the human brain and the general structure of ANNs were introduced. Then ANNs' relations with the statistical methods were investigated. ANNs' advantages and disadvantages as statistical methods were discussed. Many neural networks methods are considered generalizations of some of the classical statistical techniques. Generally, in statistics ANNs are used as flexible, nonlinear regression and classification models. Many neural network architectures have close links with the nonparametric statistical methods. Results may be obtained by training the feed forward ANNs algorithms with the nonlinear models of many statistical techniques.